Who is liable when an AI chatbot gives a customer a wrong answer, and must you say it is AI?

What did the Air Canada tribunal decide?

Jake Moffatt asked Air Canada's website chatbot about bereavement fares after a family death. The chatbot said he could submit his ticket for a reduced rate within 90 days of issue. A linked policy page said the policy did not apply to requests after travel. He relied on the chatbot, booked, and claimed. Air Canada's representative later admitted the chatbot had used misleading words¹.

The tribunal found negligent misrepresentation. It said Air Canada owed a duty of care as a service provider to a consumer and had to take reasonable care that what it told customers was accurate. It rejected the airline's suggestion that the chatbot was a separate legal entity, calling it "a remarkable submission", because the bot was part of the website and the airline is responsible for all information on it, whether it comes from a static page or a chatbot¹.

The correct policy page elsewhere on the site did not save the airline. The tribunal said Air Canada had not explained why that page was more trustworthy than its chatbot, or why customers should double-check one part of its site against another. The airline's contractual defence also failed because it did not put the relevant terms in evidence¹.

The amounts were small, the forum was a small-claims tribunal in British Columbia, and the decision binds no Indian court. It matters because of its reasoning, which others may cite.

Does the same logic apply in India?

We found no Indian court or consumer-commission ruling on a chatbot's wrong answer. The Consumer Protection Act 2019 defines deficiency in a service as any fault, imperfection or shortcoming in the quality, nature and manner of performance, and includes negligence that causes the consumer loss⁷. The definition turns on the service, not on whether a person or a program delivered it. On our reading, a chatbot answer on your site is part of your service. That is a reading, not a ruling.

Must you say it is AI?

In the EU, in most cases yes. Article 50(1) of the AI Act requires providers to design AI systems that interact directly with people so that those people are informed they are dealing with an AI, unless that is obvious to a reasonably well-informed person⁴. The Act applies from 2 August 2026². A law-firm alert says the AI Omnibus left the substance of the transparency duties alone, and gave providers of generative systems already on the market until 2 December 2026 only for the separate duty to mark synthetic output⁵.

The duty sits on the provider, the party that builds the system or places it on the market under its own name. If you build or brand your own bot you may be the provider. If you buy one, check the contract. That is our reading, so confirm it with counsel. Breaches of Article 50 can bring fines of up to EUR 15 million or 3 percent of worldwide annual turnover, whichever is higher, and for SMEs whichever is lower⁶.

In India we found no rule that makes a customer-service chatbot announce itself. The February 2026 amendment to the IT Rules covers synthetic audio, visual and audio-visual content and puts labelling duties on intermediaries. The law-firm summary we read does not mention chatbots⁸. Telling users anyway costs nothing and removes the argument that they were misled.

What does WhatsApp require of an AI bot?

WhatsApp's Business Policy allows automation when replying inside the 24-hour window, but requires prompt, clear and direct escalation paths, such as in-chat transfer to a human agent, a phone number, an email address or a support form. It also requires opt-in permission before you contact someone. Breaches can lead to limits on how many messages you can send, or to a ban from future use³.

How do you reduce the risk in practice?

• Say it is an AI in the first message.
• Give a route to a human inside the chat, and log every handover.
• Keep policy pages and bot answers consistent. Air Canada's correct page did not help it¹.
• Limit the bot to documents you own and keep current, and write down the topics it must hand over instead of guessing, such as refunds, medical or legal questions.
• Keep a record of what the bot said, so you can find and fix a wrong answer before a customer relies on it.